Fault diagnosis method and system for movable steel structure facilities
Through the three-dimensional analysis method of time domain, frequency domain and entropy value, the vibration signals of rotating steel structures are processed and the weights are dynamically adjusted, which solves the problem of insufficient noise resistance of traditional fault diagnosis methods, improves the accuracy and efficiency of fault analysis, and reduces the Safety risks brought about by improper maintenance.
Patent Information
- Application Number
- CN202411104065.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-12-31
- Estimated Expiration
- 2044-08-13
AI Technical Summary
The traditional rotary steel structure fault diagnosis method has insufficient noise resistance and is susceptible to interference from environmental factors, resulting in low analysis efficiency and accuracy, making it difficult to avoid premature or late maintenance, which may cause safety accidents.
Using a three-dimensional analysis method based on time domain, frequency domain and entropy values, the vibration signals are obtained and processed, and the matrix A1 is established, and differential analysis is performed based on motion control parameters, environmental information and historical data, and the weight is dynamically adjusted to improve the fault diagnosis. Anti-interference ability and accuracy.
The fault diagnosis of high consistency, high computing efficiency and high noise immunity is achieved, reducing the problems of poor consistency and insufficient noise immunity of traditional methods, and improving the accuracy and efficiency of fault analysis.
Smart Images
Figure CN118670661B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of steel structure maintenance, and in particular to a method and system for diagnosing faults of movable steel structure facilities. Background Art
[0002] In addition to being used to construct buildings, bridges and other engineering structures, steel structures are also used in many scenarios such as quay cranes, tower cranes, rocket launch platforms, etc. For the latter type of active large-tonnage load rotating steel structure applications, it is often necessary to periodically arrange personnel for maintenance and replacement of parts to avoid safety accidents caused by failures of the active steel structure during use.
[0003] However, traditional rotating steel structures are generally repaired by staff at fixed intervals by the relevant operators, which is greatly affected by the subjective factors of the inspectors. Too early maintenance will lead to increased maintenance time and maintenance costs, while too late maintenance may cause sudden failures and bring serious consequences.
[0004] At present, some units have proposed installing acceleration sensors on rotating parts of steel structures, installing noise sensors near transmission parts and power parts, and analyzing the failure of active steel structures by collecting vibration and noise. Although this solution can help staff to remotely analyze and discover mechanical failures, it is not noise-resistant enough and is easily affected by environmental factors. Therefore, this application proposes a new technical solution. Summary of the invention
[0005] In order to enhance the anti-interference capability of fault analysis of movable steel structures and improve analysis efficiency and accuracy, the present application provides a method and system for diagnosing faults of movable steel structure facilities.
[0006] In a first aspect, the present application provides a method for diagnosing faults of movable steel structure facilities, which adopts the following technical solution:
[0007] A method for diagnosing faults of movable steel structure facilities, comprising:
[0008] Obtaining a vibration signal collected by a monitoring device preset on the movable structure;
[0009] The vibration signal is decomposed into frequency bands, and an m*n matrix A1 is established, where m represents the frequency band and n represents the working time;
[0010] Get the motion control parameters of the active structure;
[0011] The speed v and the starting position of the current action are obtained by analyzing the motion control parameters;
[0012] Calculate the position difference based on the preset starting point position, and update n based on the speed v and the position difference, and update the matrix A1;
[0013] Assign weights to different frequency bands of matrix A1 to obtain matrix A2;
[0014] Take the features of matrix A2, perform difference analysis based on historical data, and perform fault diagnosis based on the results.
[0015] Optionally, the weights of the different frequency bands are dynamic values, and updating of the weights includes:
[0016] Obtain environmental information during each action and identify and classify it to obtain various environmental categories and corresponding environmental characteristics;
[0017] Calculate the similarity between the current action's environment information and the historical action's environment information, take the matrix A1 corresponding to the action that meets the preset similarity threshold 1, and define it as matrix A3;
[0018] Subtract matrix A1 from matrix A3 to obtain matrix B1;
[0019] The weights matched to the respective frequency bands are updated based on the matrix B1.
[0020] Optionally, also include:
[0021] If the movable structure is a cantilever structure for carrying, the weight of the heavy object being carried is obtained;
[0022] Subtract the current weight from the weight corresponding to the historical action to get the weight difference G1;
[0023] If the weight difference G1 is greater than a preset threshold, the calculation of the environmental information similarity of the corresponding historical action is stopped.
[0024] Optionally, also include:
[0025] Get the minimum failure period T1 of the component;
[0026] Take the matrix B1 corresponding to the historical actions before period T1;
[0027] Subtract the matrix A1 corresponding to the previous actions from the matrix B1 to obtain several matrices A4;
[0028] Before defining period T1, the matrix corresponding to the action whose environmental information similarity meets the preset similarity threshold 1 is A5;
[0029] Calculate the similarity between matrix A4 and matrix A5 to obtain similarity d1;
[0030] If the similarity d1 is greater than the preset similarity threshold 2, the matrix B1 is defined as a periodic pulse excitation matrix, and a firmware quantitative defect prompt is output.
[0031] Optionally, the method for obtaining the minimum failure period of the component includes:
[0032] Obtaining maintenance records of the movable structure; wherein the maintenance records include the maintenance time and maintenance method of the components;
[0033] Identify maintenance records and calculate the damage cycle of each component;
[0034] The minimum damage cycle is defined as cycle T1.
[0035] Optionally, also include:
[0036] Obtain the distance L between the transported heavy object and the installation node of the movable steel structure;
[0037] Subtract the matrix A1 corresponding to the current action from the matrix A1 corresponding to the previous action to obtain the matrix A6;
[0038] If the distance L corresponding to the current action is greater than the distance L corresponding to the previous action, and the number of negative elements in the matrix A6 is greater than the preset risk threshold, an abnormal prompt is output.
[0039] Optionally, also include:
[0040] Get the image of the active structure;
[0041] If the active structure moves, the amount of movement of the active structure is calculated based on the image to obtain the true value d1;
[0042] If the difference between the action amount and the control parameter is greater than the preset error tolerance threshold, the standard action amount of each element position in the matrix A1 is calculated;
[0043] Based on the historical images during the action process, the image time corresponding to the time when the true value d1 is the same as the standard action amount is taken, and the vibration signal features corresponding to n are updated.
[0044] In a second aspect, the present application provides a movable steel structure facility fault diagnosis system, which adopts the following technical solution:
[0045] A movable steel structure facility fault diagnosis system comprises a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute any of the movable steel structure facility fault diagnosis methods described above.
[0046] In summary, the present application includes the following beneficial technical effects: three-dimensional analysis based on time domain, frequency domain, and entropy value (complexity) can be performed on movable steel structures, and after being applied to the fault diagnosis of various types of steel structure rotating machinery, it can be used to alleviate the problems of poor consistency and insufficient noise resistance of traditional entropy value methods. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Figure 1 It is a schematic diagram of the main process of the method of this application. DETAILED DESCRIPTION
[0048] The following is combined with Figure 1 This application is described in further detail.
[0049] The embodiment of the present application discloses a method for fault diagnosis of movable steel structure facilities, which aims to: monitor the operating status of rotating machinery through time domain analysis, process periodic impact excitation through frequency domain analysis, and characterize the health status of equipment by measuring the complexity of vibration signals through entropy analysis. The method has the characteristics of high consistency, high computational efficiency and high noise resistance, which is described in detail below.
[0050] Reference Figure 1 , the fault diagnosis methods for movable steel structure facilities include:
[0051] 1) Obtain the vibration signal collected by the monitoring equipment preset on the active structure.
[0052] Among them, active structures such as the rotating cantilever of the rotating steel structure and monitoring equipment such as the vibrating string collector, their data are preferably sent back by wireless backhaul, that is, 4G / 5G data transmission terminals are deployed at the collection site to facilitate front-end information collection.
[0053] 2) Decompose the vibration signal by frequency band and establish an m*n matrix A1; where m represents the frequency band, n represents the working time, and the elements in the matrix can be amplitude values.
[0054] It is understandable that the vibration signal of the rotating steel structure has many different frequencies and oscillation components, and the specific characterizations of vibrations in different frequency bands in equipment failure are different. Therefore, this method decomposes the signal into frequency bands of different ranges and establishes matrix A1 based on them; the frequency bands can be divided into 10 equal parts in total.
[0055] Example of matrix A1: the elements of each row of the matrix represent the same frequency band, and the elements of each column of the matrix represent elements of different frequency bands at the same working time.
[0056] Therefore, matrix A1 reflects the representation of the vibration signal in the time domain, frequency domain and complexity, a three-dimensional integrated representation.
[0057] It should be noted that when the value corresponding to the meta position does not exist, it is filled with 0.
[0058] 3) Get the motion control parameters of the active structure.
[0059] It is understandable that rotating steel structures such as tower cranes must have corresponding control systems, so the above motion control parameters can be obtained by requesting and calling the matching control system; the motion control parameters in this embodiment should at least include the rotation amount / angle and rotation speed.
[0060] 4) Obtain the speed v and the starting position of the current action based on the motion control parameter analysis.
[0061] 5) Calculate the position difference based on the preset starting point position (i.e., the current starting position minus the preset starting point position, which can be understood as the angle difference), and update n based on the speed v and the position difference, and update the matrix A1.
[0062] The speed formula is known: v = distance / time, so we can get the time difference between the current action and the agreed origin when it starts at different positions, and update n accordingly to prevent analysis errors caused by time misalignment. After all, different times represent structures such as steel structure cantilevers at different positions, and the probability of large vibration when failures occur at different positions is different.
[0063] The method of updating matrix A1 is as follows: define the first column of matrix A1 with the current time and add a time stamp.
[0064] 6) Matrix A1 is weighted for different frequency bands to obtain matrix A2. For example, the weight of the first row of the matrix is 0.2, the weight of the second row of the matrix is 0.36, etc.; the specific weight coefficients are initially set manually.
[0065] 7) Take the features of matrix A2, perform difference analysis based on historical data, and output the results for fault diagnosis.
[0066] Among them, the roughened features of matrix A2 can be equivalent to the elements of matrix A2, and preferably the matrix feature vectors. After obtaining the data, one method is to compare it with the historical data one by one to complete the difference analysis, and the comparison content includes frequency, oscillation excitation size, etc.; another method is for the staff to pre-screen the historical data samples, perform 1)-6) on the selected samples and extract the features, and then train them with the neural network model, and then 7) can import the features into the corresponding model to analyze the failure rate and suspected failure characteristics.
[0067] According to the above content, this method can be used to perform three-dimensional analysis based on time domain, frequency domain, and entropy value (complexity) for movable steel structures. After being applied to the fault diagnosis of various types of steel structure rotating machinery, it can be used to alleviate the problems of poor consistency and insufficient noise resistance of traditional entropy value methods.
[0068] In one embodiment of the method, the weights of the above processes can be set equal at the initial value by the staff, and the method is automatically and dynamically adjusted based on the vibration performance of the active steel structure in daily use after application, specifically:
[0069] 1) Obtain the environmental information of each action and identify and classify it to obtain various environmental categories and corresponding environmental characteristics.
[0070] Among them, the classification of environmental information, such as temperature, weather, and wind properties.
[0071] 2) Calculate the similarity between the environmental information of the current action and the environmental information of the historical action, take the matrix A1 corresponding to the action that meets the preset similarity threshold 1, and define it as matrix A3.
[0072] The similarity calculation includes, for example, vectorizing the environmental information and then calculating the similarity using the cosine similarity algorithm.
[0073] 3) Subtract matrix A1 from matrix A3 to obtain matrix B1.
[0074] 4) Update the weights matched to each frequency band based on the matrix B1.
[0075] Among them, the weights matched by each frequency band are updated based on the matrix B1, for example:
[0076] S1. Find the ratio of the elements of matrix B1 to the elements at the same position in matrix A1 to form a ratio matrix C1.
[0077] S2. Sum multiple proportion matrices C1 and then calculate the average to obtain matrix C2.
[0078] S3, sum and average the elements of each row of matrix C2, sort them according to the size of the mean, and adjust the weight value according to the preset weight redistribution rule. For example: 5 means are sorted from large to small, and the weights are 0.5, 0.25, 0.15, 0.05, and 0.05.
[0079] Based on the above content, it can be seen that the weights in the analysis process of this method are not constant, but will be dynamically adjusted according to the actual environment and historical data.
[0080] Furthermore, because the rotating steel structure is mostly used as a cantilever of a structure such as a tower crane to transport heavy objects, and when the weights of the transported objects are different and the weight difference is large, the force conditions of the cantilever are different, which will lead to different vibrations caused by the same impact. Therefore, the method also includes:
[0081] If the movable structure is a cantilever structure for carrying, the weight of the heavy object being carried is obtained;
[0082] Subtract the current weight from the weight corresponding to the historical action to get the weight difference G1;
[0083] If the weight difference G1 is greater than a preset threshold, the calculation of the environmental information similarity of the corresponding historical action is stopped.
[0084] Among them, the weight can be obtained by the staff actively uploading it. According to the above settings, when the historical data of the active steel structure is used for analysis and weight update, it will first be screened according to the heavy objects moved to reduce the interference caused by the large force difference when the steel structure is used as a cantilever.
[0085] In another embodiment of the method, the method further comprises:
[0086] 1) Obtain the minimum component failure cycle T1. It should be noted that, considering the large number of components in the equipment, the failure cycles may be different, so the cycle T1 here should be the cycle T1 corresponding to the component that is most easily damaged.
[0087] 2) Take the matrix B1 corresponding to the historical actions before period T1, that is, the matrix B1 calculated above.
[0088] 3) Subtract the matrix A1 corresponding to the first several actions (e.g., three times) from the matrix B1 to obtain several matrices A4.
[0089] 4) Before defining period T1, the matrix corresponding to the action whose environmental information similarity meets the preset similarity threshold 1 is A5; that is, the vibration signal data under the similar environment is obtained.
[0090] 5) Calculate the similarity between matrix A4 and matrix A5 to obtain similarity d1;
[0091] If the similarity d1 is greater than the preset similarity threshold 2, the matrix B1 is defined as a periodic pulse excitation matrix, and a firmware quantitative defect prompt is output.
[0092] It is understandable that if the rotating structure has transmission gears and one of the gears is broken, the equipment will have regular periodic jitter during the transmission process. The above setting compares each vibration signal with the vibration signal when the equipment is theoretically intact, and analyzes whether the difference obtained is regular. If so, a prompt is given.
[0093] In this method, the minimum failure period T1 of the above-mentioned component is obtained by:
[0094] Obtain maintenance records of the movable structure; wherein the maintenance records include the maintenance time and maintenance method of the components, and the maintenance method includes replacement of components and repair of component defects;
[0095] Identify maintenance records and calculate the damage cycle of each component;
[0096] The minimum damage cycle is defined as cycle T1.
[0097] According to the above settings, the period T1 of this method for regular fault cause analysis is not fixed, but will be adjusted according to historical behavior; because for the same equipment, after a certain component is damaged, the staff may replace the structure of different manufacturers and brands, and the product quality of each manufacturer is different. If the value is fixed, it is easy to miss the fault occurrence node, resulting in analysis errors.
[0098] In another embodiment of the method, considering that when the movable steel structure is used as a cantilever to carry heavy objects, if the position of the heavy object relative to the steel structure changes, it will cause the center of gravity to adjust, and the long rod structure has a different center of gravity position, and the vibration amplitude and frequency of the rod or arm are different when vibration occurs, such as: the farther away from the fixed position, the greater the difference, so this method is set as follows:
[0099] Obtain the distance L between the heavy object being transported and the installation node of the movable steel structure (such as the main pole of the tower of the steel structure cantilever). This data can be obtained from the control system of the steel structure facility;
[0100] Subtract the matrix A1 corresponding to the current action from the matrix A1 corresponding to the previous action to obtain the matrix A6;
[0101] If the distance L corresponding to the current action is greater than the distance L corresponding to the previous action, and the number of negative elements in matrix A6 is greater than the preset risk threshold, that is, the vibration characteristics of multiple frequency bands and times are smaller than the vibration characteristics when the center of gravity is close to the inside, then an abnormal prompt is output.
[0102] In another embodiment of the method, an important premise for the implementation of the above content is that the actual action amount and the control amount of the structure are the same, and equipment wear, failure, and thermal coefficient interference will cause errors in the distribution of elements in the matrix A1, which will lead to analysis errors. Therefore, this method is set as follows:
[0103] Get the image of the active structure;
[0104] If the active structure moves, the amount of movement of the active structure is calculated based on the image to obtain the true value d1;
[0105] If the difference between the action amount and the control parameter is greater than the preset error tolerance threshold, the standard action amount of each element position in the matrix A1 is calculated;
[0106] Based on the historical images during the action process, the image time corresponding to the time when the true value d1 is the same as the standard action amount is taken, and the vibration signal features corresponding to n are updated.
[0107] Among them, calculating the movement amount of the same feature based on different images is an existing technology, so it will not be repeated. That is, if the actual movement amount of the structure is quite different from the standard value, the standard movement amount corresponding to each element in the matrix A1 is calculated to determine the time when the same movement amount actually occurs, and the vibration characteristics at that time are taken to fill the position of the corresponding element in the matrix A1, so as to reduce the interference caused by the difference between the control parameters and the actual movement of the active structure and improve the accuracy of the analysis.
[0108] The embodiment of the present application also discloses a movable steel structure facility fault diagnosis system.
[0109] A movable steel structure facility fault diagnosis system, the movable steel structure facility fault diagnosis system comprises a memory and a processor, the memory stores a computer program that can be loaded by the processor and executed by any of the movable steel structure facility fault diagnosis methods described above.
[0110] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.
Claims
1. A method for diagnosing faults of movable steel structure facilities, characterized in that: include: Obtaining a vibration signal collected by a monitoring device preset on the movable structure; The vibration signal is decomposed into frequency bands and an m*n matrix A1 is established, where m represents the frequency band and n represents the working time; Get the motion control parameters of the active structure; The speed v and the starting position of the current action are obtained by analyzing the motion control parameters; Calculate the position difference based on the preset starting point position, and update n based on the speed v and the position difference, and update the matrix A1; Assign weights to different frequency bands of matrix A1 to obtain matrix A2; The features of matrix A2 are taken, and difference analysis is performed based on historical data, and fault diagnosis is performed according to the results; the weights of different frequency bands are dynamic values, and the updating of the weights includes: Obtain environmental information during each action and identify and classify it to obtain various environmental categories and corresponding environmental characteristics; Calculate the similarity between the current action's environment information and the historical action's environment information, take the matrix A1 corresponding to the action that meets the preset similarity threshold 1, and define it as matrix A3; Subtract matrix A1 from matrix A3 to obtain matrix B1; The weights matched to the respective frequency bands are updated based on the matrix B1.
2. The method for diagnosing faults of movable steel structure facilities according to claim 1, characterized in that: Also includes: If the movable structure is a cantilever structure for carrying, the weight of the heavy object being carried is obtained; Subtract the current weight from the weight corresponding to the historical action to get the weight difference G1; If the weight difference G1 is greater than a preset threshold, the calculation of the environmental information similarity of the corresponding historical action is stopped.
3. The method for diagnosing faults of movable steel structure facilities according to claim 2, characterized in that: Also includes: Get the minimum failure period T1 of the component; Take the matrix B1 corresponding to the historical actions before period T1; Subtract the matrix A1 corresponding to the previous actions from the matrix B1 to obtain several matrices A4; Before defining period T1, the matrix corresponding to the action whose environmental information similarity meets the preset similarity threshold 1 is A5; Calculate the similarity between matrix A4 and matrix A5 to obtain similarity d1; If the similarity d1 is greater than the preset similarity threshold 2, the matrix B1 is defined as a periodic pulse excitation matrix, and a firmware quantitative defect prompt is output.
4. The method for diagnosing faults of movable steel structure facilities according to claim 3, characterized in that: The method for obtaining the minimum failure period of the component includes: Obtaining maintenance records of the movable structure; wherein the maintenance records include the maintenance time and maintenance method of the components; Identify maintenance records and calculate the damage cycle of each component; The minimum damage cycle is defined as cycle T1.
5. The method for diagnosing faults of movable steel structure facilities according to claim 4, characterized in that: Also includes: Obtain the distance L between the transported heavy object and the installation node of the movable steel structure; Subtract the matrix A1 corresponding to the current action from the matrix A1 corresponding to the previous action to obtain the matrix A6; If the distance L corresponding to the current action is greater than the distance L corresponding to the previous action, and the number of negative elements in the matrix A6 is greater than the preset risk threshold, an abnormal prompt is output.
6. The method for diagnosing faults of movable steel structure facilities according to claim 5, characterized in that: Also includes: Get the image of the active structure; If the active structure moves, the amount of movement of the active structure is calculated based on the image to obtain the true value d1; If the difference between the action amount and the control parameter is greater than the preset error tolerance threshold, the standard action amount of each element position in the matrix A1 is calculated; Based on the historical images during the action process, the image time corresponding to the time when the true value d1 is the same as the standard action amount is taken, and the vibration signal features corresponding to n are updated.
7. A fault diagnosis system for movable steel structure facilities, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executes a method for diagnosing faults of movable steel structure facilities as claimed in any one of claims 1 to 6.